unexpected NOP encodings

[Peter Bergner]

On 6/24/19 10:05 AM, Alan Modra wrote:
> On Mon, Jun 24, 2019 at 07:56:48AM -0600, Jan Beulich wrote:
>> All,
>>
>> is it known (and intended) that
> 
> No, you've found a bug.

Agreed, the group ending nop generated by the .p2align should match
the .machine that is currently in effect, so power8, not power6.

Peter


>> 	.text
>> 	.machine power8
>> power8:
>> 	nop
>> 	# The NOP inserted here is controlled by the "power6" machine
>> 	# setting below, not the one currently in effect.
>> 	.p2align 3
>> 	blr
>>
>> 	.machine power6
>>
>> doesn't produce the expected NOP encoding, because the encoding is
>> chosen only at the end of the assembly process (i.e. when the final
>> directive has already been seen and processed)?
>>
>> Jan
>>
>